Towards precise classification of cancers based on robust gene functional expression profiles.

Area of Science:

• Computational biology
• Bioinformatics
• Genomics

Background:

• Analyzing high-dimensional gene expression profiles is crucial in computational biology.
• Genes function in a modular fashion within cells, suggesting a need for module-level analysis.
• Current methods lack precise classification of complex human diseases at the molecular level.

Purpose of the Study:

• To develop a computational algorithm for precise disease classification using functional gene expression profiles.
• To reduce data dimensionality by representing gene expression at the functional module level.
• To improve the interpretation of molecular mechanisms in complex human diseases.

Main Methods:

• Annotated individual genes to functional categories using Gene Ontology.
• Identified functional categories enriched with differentially expressed genes.
• Computed summary measures for gene expression within functional modules to represent module activity.
• Compared classification performance of functional versus conventional gene expression profiles using decision trees.

Main Results:

• Developed a low-dimension functional expression profile for data reduction.
• Functional expression profiles enabled precise classification of tumor types across four datasets.
• The modular approach improved the interpretation of complex biological data.

Conclusions:

• The modular approach offers a robust alternative for analyzing high-dimensional microarray data.
• This method is resilient to high measurement noise and biological variance.
• It facilitates efficient integration with existing biological knowledge for disease mechanism interpretation.